COMMENT ⊗   VALID 00007 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	paradi[f88,jmc]		Notes on Paradigms Lost by John Casti
C00010 00003	Chapter 4
C00016 00004	skipping ahead
C00049 00005	Chapter 6  Where are they?
C00056 00006	Chapter 7 How Real is the Real World
C00059 00007	Notes:
C00060 ENDMK
C⊗;
paradi[f88,jmc]		Notes on Paradigms Lost by John Casti

still quite a few misprints and spelling mistakes, e.g. tantalyzing
exhuberance, quandry, monkies, about we Earthlings

Is there an ``essence of man''?  It seems to me that the
assumption that there is isn't essential to the content
of the book.

49 Goedel didn't fall for the instrumentalism of the
Vienna circle.  In fact his succeses were based in
a large measure on his philosophical platonism.
Both his PhD thesis on the completeness of first order
logic and his more famous incompleteness theorems
are based on making the distinction between provability
and truth and examining when they happen to coincide.

55 The problem of continuing sequences can be made more sensible
by using Chaitin-Kolmogorov complexity.  However, it would be
interesting to understand why so many people agree on
what is the natural continuation of a sequence.  This might
involve criteria for what continuations are most compactly
described by a human.  Perhaps it would depend on how much
mathematics the human knows.  There is, of course, the famous
example of continuing the sequence (14 34 42 59).  Almost
everyone who has a definite opinion chooses 125.

59 Lakatos took an untypical problem, Euler's theorem, in his
proofs and refutations.  It was in a domain, topology, that no-one
attempted to make rigorous until the 20th century.

69 Every graduate student now defines his own paradigm shift,
and Vaughan Pratt had 5 paradigms in one paper.
I sometimes conjecture that Kuhn, Lakatos, Feyerabend, etc.
have corrupted part of a younger generation of scientist.
People who have been taught that new scientific views are
largely a matter of propaganda sometimes feel themselves an
obligation to turn themselves into propagandists.  Connectionism
has developed in an aura of propaganda.  Much of the propaganda
is just ``This is the wave of the future; all else is old-fashioned.''

78 Einstein succeeded in convincing Ostwald of the reality of
atoms.  The history here seems dubious.  Maxwell's kinetic theory
was based on atoms and molecules, and he died in 1879.

81 Leave aside the question of whether a particular scientist is
sufficiently detached.  It can only lead to inconclusive recrimination.
Rely on the public objective process to correct errors.

94 I wasn't convinced by the criteria for distinguishing pseudo-science.
Wegener's notion of continental drift based on correlations between
the geology of Africa and that of South America had a tremendous gap ---
the lack of a proposed mechanism.  This gap caused it to be dismissed
until Wilson thought of a mechanism around 1960.  Nevertheless, it wasn't
pseudo-science.  There's a quantitative element here.  Velikovsky
required not just one but many arbitrary exceptions to current scientific
doctrine.

97 The notion of religion given here is a cleaned up version designed
to avoid conflict with science.  Its only continuity with pre-scientific
religion is institutional.  Pre-scientific religion really tried to
answer scientific questions.  Given the knowledge the ancients had,
the idea that thunder was the result of a god's hammer blows was
plausible.  However, there was no scientific method for testing it
and no better rival theory.

101 The controversy about atoms lasted a long time but was settled.

102 Scientism proposes that science is the only way of getting such
truth as may be had.

Chapter 2

103 1953 also gave us the first four minute mile, the conquest of Everest,
the East German workers' revolt and the coronation of Elizabeth II.

108 I suppose Miller is still alive and could be telephoned to as
what he knew about Calvin's work.

123 Dates and references on Spiegelman and Eigen would be nice.

166 Casti is gullible about stories that ascribe ignorance.
In fact the Indiana Legislature referred the proposal to set pi
at 3 to the Committee on Swamps which didn't report out a bill.
Alas, the stereotype of a Middle Western Legislature proved more
enduring than the afore-mentioned Middle Western Legislature's
humor.

Chapter 3

201 To say that a trait is not adaptive or maladaptive in itself
but only relative to an environment can be misleading.  Suppose
we contemplate giving a randomly chosen person a million dollars.
Of course, we can concoct a hypothetical environment in which
this would be harmful, but suppose there is no actual person
in the world in that situation at present.

My objection is that one should not always honor a request to
take one more factor into account.  My guess is that many
qualities are maladaptive in every environment to which the
individuals of a species exist in.  For example, Down's syndrome
is probably maladaptive for humans in every actual environment.
It is harder to come up with adaptive traits that clearly apply
to all environments.  However, it takes an argument to show that
most adaptive traits are of limited use.  Of course, white
color in a bear is an example.


Basically I have no quarrel with Chapter 3
Chapter 4

277 I think Clever Hans has been overworked.

278 I'm not sure Premack, et. al. would agree with your bottom line.

282 Doesn't Chomsky's book on Cartesian linguistics recognize
17th century precursors of his ideas?

284 It seems to me that the ``poverty of the stimulus'' argument
is weak.  It depends on assumptions, which may be false, about
what kinds of structures can be recognized in examples.

283 I'm not sure children learn languages more easily than adults.
The language used by a three year old has a small vocabulary and
simplified grammar.  Perhaps an adult could learn what a two year
old learns even easier.  Suppose I spent each day observing
through one way glass a Hungarian speaking mother and her two
year old child, knowing no Hungarian initially.  Perhaps after
a year I would know as much childish Hungarian as the child.
People who change linguistic environments as a child often discover
that all they know of their original language is childish.

283 Figure 1 doesn't correspond to the text.  It would be better
to have an example that doesn't require the reader to know about even
the most elementary electric circuits.

284 There is nothing about probabilities in the definition of
finite state grammar, and your examples don't actually involve them.

295 You mean ``generate sentences of arbitrary finite length'' in the
second paragraph - not infinite length.  There's always a reachable
stop state in the finite state grammars that are used.  I didn't check
it out in your example.

298 It seems to me that phrase structure grammars allow sentences
to be generated in more than one way.  There needn't be a unique
rule for each non-terminal.

300 ``the semantic component of the language's grammar'' is not
previously explained.

318 Might the Nootka phrase also mean, ``He forced them to eat''.

329 Politically, Chomsky is allied to the people who attack Wilson.

329 Wasn't Piaget over 80?

335 I'm not sure Fodor would consider himself in harmony with Chomsky.

Two final points.

1. I think neither Chomsky or any other linguist gives semantics
its due.  Language exists for communication, and they almost
entirely ignore studying what one person might know and have to
communicate to another and how this depends on what he can
presume about what the other already knows.  I have faced this
problem somewhat in thinking about a ``Common Business
Communication Language'' that would communicate among computers
belonging to different businesses inquiries about what products
the receiving business markets with given specifications.  The
reply must also not presume too much about what the questioner
knows.  I believe that theories of language will be successful
only when they make this problem central.  It will not even be
approached just by determining what sentences are grammatical.
Understanding how children acquire language also depends on
understanding what is communicated by and too children.

2. In this and the succeeding chapter you ignore Newell, whose
influence on psychology has been revolutionary.  It was the AI
approach, and Newell in particular, that really clobbered behaviorism.

skipping ahead

340 Raising the issue ``Can a computer think?'' skips lightly over
the question of what we can assert that machines definitely can
do.  It is like entering biology with the question ``Can we create
an artificial mouse?''  These questions can be argued, and the
latest AI or biology is relevant to the discussion, but in neither
case is the understanding of a layman or beginner greatly advanced
by starting with it.  You get to the highest level issues immediately
without ever bothering with the scientific detail of what has actually
been discovered.

340 I see Casti accepts uncritically Weizenbaum's picture of the computer
hacker.

341 Standard politics, I see.

343 The characterization of Turing is guesswork and probably wrong.
Why shouldn't Turing's interest in AI have been derived from his
prewar work in making the notion of computable number precise ---
the work that led to the universal Turing machine.  Unfortunately,
Turing has had to serve as a character in various people's morality
plays.  Casti's too.

343 bottom ``asserting that the existence of ``thinking'' is solely
a matter of producing convincing responses to more or less arbitrary
stimuli'' is incorrect.  Even if one accepts the imitation game
as giving a sufficient condition for saying that a machine thinks,
this doesn't commit one to the ideas that the existence of thinking
is the same as playing the imitation game.  My opinion is that
Turing's motivation for the imitation game was to smoke out
philosophical views that would accept no behavioral evidence for
thinking.  It did not commit Turing a behavioral notion of thinking,
and it seems to me that such a notion is not exactly in the
spirit of Turing's earlier work.  Neither the Entscheidungproblem
paper nor his work on hierarchies of computable ordinals suggest
a fundamentally behaviorist orientation.

346 See my commentary on the Chinese Room associated with Searle's
article in BBS.  There I point out that we have an example of one
system (the human) interpreting another (the Chinese conversationalist).
This is common in computing, and it is clear that one doesn't ascribe
the the basic machine the abilities of the program it is interpreting.
A time-sharing computer is executing all sorts of different programs
with different capabilities.  Searle's intuition pump, to use
Dennett's terminology, also slips by the fact that a human simulating
rules for Chinese conversation as described might be too slow by
a factor of 100 million.  One might be doubtful about the possibility
of animated movies if one imagined that the Disney artist drew each
frame and then displayed it to the audience.

346 Casti is using ``society of mind'' in a quite different sense
than Minsky, who coined the phrase.

348 Casti's use of ``formal logical system'' is correct, but he
should be careful that it doesn't necessarily involve the use
of sentences of formal logic to express facts.  Let's see if he
slithers from one to the other without adequate notice.

He should also be careful about the meaning of ``physically executing
the rules called for by the rules of a formal logical system''.  In
general, the rules of a formal logical system don't call for
executing anything.  There may be programs for testing grammaticality
of strings, programs for testing the correctness of single inferences
and an infinite variety of programs for trying to prove theorems
or to find strings characterized by sentences.  Betcha he muddles
them.

The term ``rules of inference'' is used in a nonstandard way.  Casti
uses is as if it were the rules that build complex grammatical strings
from simpler ones.  There isn't an absolutely standard terminology
for this, but ``rules of formation'' is a term commonly used.  The
term ``rules of inference'' applies in the special case when the
grammatical strings are interpreted as sentences and the ``rules''
determine when one sentence is considered to be immediately
inferrable from axioms and previously inferred  sentences.
Neither chess nor programming languages have ``rules of inference''
in the standard sense of the term.

349 In the chess example, it is uninformative to say that the rules of
inference consist of the legal moves, although the sentence can
be interpreted in such a way that it is correct.  Normally, if one
wanted to interpret rules of chess as a language, one would want
a small number of rules of inference, not much larger than the
number of sentences in a ``How to play chess book''.  The number
of legal moves taking into account the varied positions is enormously
larger.  The rules of chess can be used as the basis for a wide
variety of formal systems.  Many uses require a formal system
that tells who won the game --- something Casti omits.

349 Politics to the front, as the Maoists used to say.

350 The description of scrabble as a formal system is sloppy.
If the strings are taken as words, a scrabble position is not
characterized by the set of words on the board.  Hence whether
a new word is admissible is not given by rules of inference
in the manner defined previously.  To make scrabble into a
formal system, it is necessary to take scrabble positions
as strings.  It is also necessary to decide whether to include
in a position the set of letters in possession of a player
and how to handle the set of face down letters.

The author might suppose that asking that this be done correctly
is mere pedantry, but a naive reader will not get a correct
notion of the well-definedness of formal systems if those proposed
by the author as examples are not actual formal systems.

351 The general remark about the fascination of games for AI
researchers is misleading.  The fascination comes from the
possibility of programming a strategy, and this possibility is
omitted from the formal systems given as examples, if only
because the criteria for winning are omitted.

353 The ``addition example'' is doesn't have the claimed interpretation.
The axioms only generate strings consisting of a string of 0s
followed by a string of *s.  These aren't addition sentences.
To get them requires a more sophisticated formal system.  The
rules correspond only to adding 0 and 1 and tell us nothing
about adding two arbitrary natural numbers.  By the way, the
mathematicians include negative numbers in the integers and
say ``natural numbers'' when the only want the non-negative
ones.

354 This page contains the common, e.g. from Quine on, philosophical
error of radical translation.  I discuss this error briefly in
my ``Ascribing Mental Qualities to Machines''.  Philosophers
sometimes say that a given chemistry book in English might be a
novel in some other language.  Any cryptographer will tell you
that this is false about any language a human could speak or
write.  For example, the English and Russian languages have
20 letters in common including a selection of vowels and
consonants in both languages.  Namely, the longest ambiguous
text anyone I know has been able to find is three letters.  An
example is POT.  The Russian word that looks like that is
pronounced  rot  and means ``company'' (in the military sense).
Shannon's discussion of the mathematics of cryptography essentially
covers this point.  The Chaitin-Kolmogorov complexity theory
is also relevant.  Take your English chemistry book.  The description
in English or mathematics of the simplest language in which
it can be interpreted as a novel would be immense if it covered
much else besides this text.  If we only want this text we can
use the following rule.  If the text is the second edition
of Linus Pauling's ``The Nature of the Chemical Bond'' then
it is to be translated as the first edition text of Harriet Beecher Stowe's
``Uncle Tom's Cabin''.  All other texts are identical in
the two languages, including other editions of Pauling's
book, including those in which a single letter is changed.

The import of the matter is that the syntax can practically
determine the semantics in that there is only one admissible
concise semantics.  If we add to the text we are considering,
some description of the circumstances in which it is uttered,
we get a more elaborate language.  That this language has
an essentially unique simple interpretation is partly responsible
for the fact that children can learn to understand and speak
their native languages.

356 The definition of completeness is wrong.  Completeness is not
a purely syntactic notion.  It is generally taken relative to
a semantics.  For example, the completeness of first order logic
is defined by the fact that all sentences ``true in all interpretations''
are theorems, and the incompleteness of arithmetic is an abbreviation
for the fact that no consistent formal system for arithmetic
can be complete in the sense that all true sentences of arithmetic
are theorems.  ``Interpretation'' and ``true'' are semantic notions.

356 The definition of consistency is one of those often given but
it depends on the presence of a negation symbol or operator in
the language, so that  not-p  is defined when  p  is.  None of the
formal systems you have defined has such a symbol or operator.
The example attempted for chess is wrong, because you haven't
indicated any notion of the negation of a position.

358 The summary table is incorrect.  The notion of rule of inference
mentioned gives only formal systems in which each sentence
follows from exactly one previous sentence.  Your examples, in so
far as they are well defined have this property, but it doesn't even
allow for modus ponens --- from  ``p''  and  ``p imples q'' infer
``q''.  Such systems are known to be limited in their capabilities.
I haven't looked for other errors in the table.  Oh well, here's
another.  If the language contains negation as Casti seems to
assume, every complete system in Casti's sense is inconsistent in
his sense.

359 The definition of computer doesn't provide for branch instructions.
Without them there is nothing useful can be done with the ability to
determine whether one number is bigger than another.

365 Chaitin came from Argentina, but he may well consider himself
an American.  Perhaps you should ask him.

367 Hilbert wasn't convinced that Goedel's work had destroyed his program.

368 It might be worthwhile to mention Goedel's arithmetization of
mathematical assertions, so that sentences about provability became
sentences about numbers.  The details are indeed difficult, but it
can be stated that he showed that there is a sentence about numbers
that in this interpretation asserts its own unprovability.

370 Surely the first sentence is incorrect.  Memory addresses often
hold more than one byte.

371 It's also but gleam in the eyes of neurophysiologists, whose actual
business it is.

372 Goedel didn't say that the human is a universal truth machine.

373 With all the sloppiness of the previous pages, the summary on this
page is essentially correct.  However, a reader who thinks he has followed
the argument is deluded and will have had his credulity enhanced, so
that he will be convinced in the future by sloppy arguments.  In particular,
whether he believes  p  or  not-p  will depend on which charlatan
gets to him first.

374 Already at Dartmouth, I was think about thinking in terms of
problem solving ability rather than simulation of human behavior.
So was Arthur Samuel who was already putting in his checker
learning program position evaluation functions that were
explicitly not based on any human idea, with the idea that maybe
there were readily computed ways of evaluating checker positions
that wouldn't occur to any human.  Newell and Simon took were
definitely committed to studying the way humans do it.

By the way, I dissolved the Stanford Artificial Intelligence Laboratory
in 1980, because the subgroups could get their own grant support.
I only formed it in order to justify a big enough computer, and
by 1980 that was no longer a problem.

375 Strong AI and weak AI are terms coined by Searle.  No-one else
purports to understand what he means by them.

375 Searle is imprecise, and I'll bet Gunderson is also.

376 The second paragraph is mistaken unless Casti believes that all
mathematical logic including Goedel is devoid of scientific
interest.  It is also presumptuous about the motivations of people
whose opinions differ from his.

377 Given the sloppiness of the notions of computer and formal system,
the two versions of ``can machines think'' have no clear meaning.
The question can be made meaningful in other ways.

378 It isn't clear that von Neumann did any work on AI.  He worked
on computers, and he worked on the resemblances between the nervous
system and electronic digital machines, and he worked peripheral
topics like making reliable machines out of unreliable components
and making self-reproducing machines.  However, he published nothing
on the possibility of artificial intelligence.  I asked his friend
Stanislaw Ulam what von Neumman thought about the chess program
Ulam collaborated on.  After a moment's thought Ulam replied, ``I
never mentioned it to him.  I didn't think he would be interested.''

381 There is an auxiliary ``top-down thesis'' that the problems
human thought processes solve can be solved by symbol processing.
This is uninteresting to Casti and Searle et. al., because they
understand neither the problems that have arisen nor the extent
to which they have been solved.  They resemble someone who said
that the biology of digestion is uninteresting except in so far
as it helps settle the problem of whether artificial life is
possible.

382 The participants in the Dartmouth study were indeed impressed
by the logic theory machine.  At least I was.  However, I was
unconvinced by the contention that they had really simulated very
much of the mental processes of a person proving theorems
expressed in logic.  I think being unconvinced on this score may
have been part of Minsky's motivation for proposing a geometry
theorem prover that summer.  I telephoned Minsky, and he says he
doesn't remember very well, but it may have been part of his
motivation.  He also said he was impressed by the Newell-Simon
work.

The geometry theorem prover, as subsequently implemented at IBM,
used a simulated diagram and only attempted to prove sentences
that were true in the diagram.  This is a special case of using
examples.  The logic theory machine was purely syntactic and
often tried to prove sentences that were false in examples.
Nevertheless, we were impressed because their work was the first
approach to systematic symbolic computation in AI.  I took the
idea of list processing from their description of IPL, combined
it with ideas from Fortran and added some new ideas to develop
LISP several years later.  Who said we were unimpressed?

They considered logical theorem proving simply as a domain for
AI rather than using logical sentences to express the beliefs
of a program about its problem domain.  I introduced this in
1958.

383 My criticism of GPS and its successors is that the idea
of state differences is too unitary to provide a good mechanism
for general problem solving or to model how humans solve problems.

384 Many people including me agree with Bar-Hillel's claim that
an encyclopedia of common sense knowledge is required to understand
language.  We propose to provide such an encyclopedia.  Lenat
proposes to plunge in and do it.  I and many others think we
need to understand better how general common sense facts can
be represented in such an encyclopedia before starting to build it.
Otherwise, the sentences will not be general enough, and an
excessive number of sentences would be required.

399 While I am not involved in making expert systems, I don't understand
the basis for Casti referring to ``expert system peddlers of
the Feigenbaum school''.  It seems that some expert systems
are useful.  Moreover, there is an AI doctrine associated with
the Feigenbaum school.  It is that all intelligence requires is
a very large number of rules like those in MYCIN.  I don't agree
with it, but it isn't evident, especially at Casti's level
of understanding, that the doctrine is wrong.

Feigenbaum's belief that AI is just a matter of a large enough
number of rules led to the creation of the AI industry.  This
industry has had a few successes, but I think that many of its
goals require more sophisticated knowledge representation than
the present AI shells provide.

399 Lenat hasn't been ``of Stanford'' for some years.  He has
been ``of MCC in Austin, Texas''.

402 Lenat's work doesn't exactly count as bottom-up in the sense
earlier mentioned in the chapter.  It's a mixture.

405 There is no special connection between connectionism as a
doctrine and the Connection Machine.  Most of the buyers of
Connection Machine (maybe all) use it for numerical computation,
e.g. for solving partial differential equations by basically
straightforward numerical methods.

406 Hinton moved to Toronto some time ago.
Rumelhart moved to Stanford a year ago.

413 Armer hasn't been at Stanford for more than five years.

419 The Lucas argument is like my saying to you, ``Let's see who
can name the largest number.  You go first.''  The process
of generating a sentence unprovable in a given
axiomatic theory of arithmetic is purely mechanical and
can be carried out by machine.  Consequently, the machine
can say to Lucas, ``Tell me the theory of arithmetic you
use and assure me that it is consistent and I'll give
you a true sentence you can't prove.''  Turing discussed
transfinite hierarchies of theories arithmetic.  Each
theory was obtained from collections of earlier theories
in the hierarchy by the Goedel theorem and by taking unions
of earlier theories.  The matter was further elaborated
by Feferman, but it's too technical for this discussion.
I consider the technicalities relevant to arguments in
the Lucas style, although I don't understand them myself.

424 More propagandist garbage about ``today's go-go expert system
propagandists'' from someone who makes no attempt to evaluate
specific claims.  Just another would-be participant in the lynch
mob.

425 I believe Weizenbaum preceded Colby in this area.  That's
one of the reasons Weizenbaum so dislikes Colby.  Weizenbaum
now says he regarded Eliza as a mere programming exercise
(in his language SLIP which sank without a trace).  However,
read his 1964 paper in CACM about Eliza, and you will see that
he regarded it then a precursor of greater things to come.

427 The story about the Russian is probably false.

427 I'm not sure what ``greeted with outrage'' means.  It was
itself an expression of outrage.  I think we gave as good as
we got.  I'm not nearly as annoyed with Weizenbaum as I am
at fools like Victor Weisskopf who certified Weizenbaum
as a computer scientist with the arrogance and ignorance
of a nuclear physicist turned bureaucrat.

430 I never heard of Rucker.

438 The essence of using the information twice, once to do something
and once to copy itself is already present in Church's? 1929?
Y combinator, written in lambda calculus notation
(lambda (f) (lambda (x) f(x(x)))((lambda (x) f(x(x))))).
When this combinator is applied to a function  F,  it generates
a fixpoint of  F.  The subexpression  x(x)  is what does the
magic of using the same information in two different ways.
Von Neumann surely new about this, and probably considered
it so standard by the late 1940s that he didn't even have
to give specific credit for it.  I remember some dumb philosopher
of the early 50s wrote a paper proving that self-reproducing
machines were impossible, because you needed a description of
the description and a description of that and so on in an
infinite regress.

440 Casti is entitled to his intuition that top-down doesn't have
``the right feel'', but he should realize that his understanding
of what different people, e.g. me, are trying to do is extremely
limited.  The whole chapter says nothing about the use of
mathematical logical languages to express what a program knows
about the world, and a fortiori it doesn't mention the development
of formalized nonmonotonic reasoning, the most significant
theoretical development in the area.  It began in the late
1970s.

444 I am less optimistic than Casti about genuine machine intelligence
being available within the next decade or two.  But then I
know about more hard problems whose solutions seem to be
to be necessary and on which hardly anybody is working.


Chapter 6  Where are they?

443 If after another hundred years - not to speak of a few million -
it still seems reasonable to communicate at 1420 megahertz, Project
Ozma will seem to have been a good idea.

452 The assumption that societies will spend a limited number
of years in a communicative mood is the most dubious of
Drake's assumptions.  Some may, but others may keep chattering
indefinitely.

472 Thanks for moving on.

474 What about interstellar expansion?  Some of us will surely do it.

481 The following anecdote suggests that Dyson needn't have been so
delicate about the Russian word filosofiya.  In 1968 I was asked
to give a lecture at the Institute of Problems of Control in
Moscow.  When I suggested to my host the title ``Philosophical
Problems of Artificial Intelligence'', he proposed ``Recent Work
on AI at Stanford'' with the remark that the original title would
bring ``the wrong people'' to the lecture.  Oh, well, I said.  I
was surprised, however, that when he introduced me he told the
story of my original title and his worry that the wrong people
would come.  This got a laugh from the audience.  I am inclined
to draw the moral is that Dyson is too inclined to suppose that
the official Soviet view is the one held by most Soviet
scientists and other people.  His book Weapons and Hope also
suggests this oversimplification.

In December I'm giving a series of lectures with the title
Philosophy and AI at the Institute of Philosophy in Moscow.
We'll see if there are any Marxists left.  They seemed to be
rather isolated at the International Congress on Logic, Methodology
and Philosophy of Science that was held in Moscow in August 1987.
The Marxist sessions were almost entirely in Russian without
translation, and I received several compliments from Soviets when
I asked a question at one of them.

481 Whereas the Germans are rational about automobiles?  We kill
fewer per mile driven than any Europeans.

508 The argument for the infeasibilty of interstellar travel is
wrong.  Suppose you use a nuclear reactor of the kind used on
submarines to expel a working fluid with an ion rocket of the kind
NASA is currently though sluggishly developing.  Such a system could
reach the nearest stars in the low thousands of years.  So you need
a multi-generation journey unless you wait till humans live much
longer.  No-one would be motivated to start now, unless the loser
in a global contest wants to send an expedition of true believers
to flee the evil that is conquering earth.  Probably either side
in WWII would have launched such an expedition if doomed to lose
and if it had the technology.  However, if humanity lasts for
hundreds of millions of years on this planet, some of us will
surely undertake journeys of even hundreds of thousands of years.

There is a formula for the time to make a journey, assuming non-relativistic
velocities.

  t = 2 s↑(2/3)p↑(-1/3)

Here  s  is the distance and  p  is the power per unit mass of the
rocket system, i.e. a figure of merit.  The exhaust velocity is
assumed to be optimally varied during the journey.  If you assume
MKS units, t is in seconds, s is in meters and  p  in watts per kilogram.
The number 2 is an approximation.

There are other schemes.  For example, Robert Forward has a neat
scheme in which the power to expel the mass is transmitted by laser
from solar orbit.  He has done the physics correctly.

523 I think you are too cynical about the motivations of scientists -
and probably about the motivations of Congressmen also.

527 We would be interested in aliens with quite different motivations
than ours.  Why wouldn't they be interested in us on account of the
curiosity that they had to have in order to develop the science
that is the basis of a technology capable of interstellar communication.

You should look at an old book called ``Lincos: a Language for Cosmic
Intercourse'' by the Dutch mathematician Hans Freudenthal.  It
discusses what to transmit to establish communication.  I think
Rescher would have a hard time arguing that Freudenthal's methods
wouldn't work.  It was written in the 50s or earlier.

Chapter 7 How Real is the Real World

549 That the electrons arrive as particles is not exactly what
quantum mechanics says.  If they interact with molecules of
phosphor in such a way that a molecule is left in an observably
altered state, that is correct.  However, consider the following
variation of the photo-electric effect as treated by Planck and
Einstein.  Use a cesium surface with strips of insulator on
it.  Let the width of the strips be as small as feasible and
also let the intervals between them be small, i.e. make a grating.
If the photon ejects an electron from an atom, leaving the
atom ionized, the behavior will correspond to the photon arriving
at a specific place.  However, cesium is a metal, and conduction
electrons in a metal are not localized.  If a conduction electron
is emitted by a non-localized photon, then it needn't be emitted
at a specific location.  If the coherence length of the conduction
electron is many times the period ff the strips, an emitted
electron will interfere with itself, and an array of electron
detectors will observe an interference pattern.

Most physicists I have asked about this, agree that this is what
quantum mechanics calls for.  The Stanford expert on
the photoelectric effect, William Spicer, thinks that it would
be a difficult but possible experiment.  I'd like to try it, but
I'm pretty far from being capable of doing experimental physics,
even though a friend at the Stanford Linear Accelerator Center
has offered some help from his group in designing and building
the apparatus.

The thinking that led to the proposed experiment is along
Everett's lines, and some future variation of that is what
I would bet on.

625 Alas, the late Heinz Pagels.

627 You mean nanonano nano second not giga.

So, you also came to the Everett view.
Notes:
postface 637
ch 7, how real is the real world, 538
ch 6, where are they?, 446
ch 4, speaking for myself, 275
intro, 1
contents, 15
ch 1, faith hope and asperity, 16
ch 2, warm little pond, 103

General comments:

The essence of man is taken as rather passive.  There is nothing about
what man will do or might do, either individually or collectively.